Jeff Davis <pgsql@j-davis.com> writes:
> On Mon, 2010-08-30 at 14:05 -0700, Alex Zepeda wrote:
>> I dumped a table with an exclude constraint with pg_dump, and loaded it
>> into another instance of postgres where it promptly failed to create the
>> constraint. The complaint was... key conflicts with key. IOW,
>> duplicate data.
> If you dump a floating-point value, and then reload it, it may be
> different than the one you started with.
That's really *not* supposed to happen, assuming that both machines have
IEEE float arithmetic and competently written float I/O code.
My own WAG is that we're talking about a GiST bug that causes it to
recognize or not recognize duplicates depending on order of insertion.
But that's theorizing far in advance of the data.
> If you are not using floating point values, please try to make a
> self-contained test case that includes data that can reproduce the
> problem.
Yeah, a test case would be awfully helpful here.
regards, tom lane